The FASEB Journal express article10.1096/fj.02-0689fje. Published online May 20, 2003.
Tetracycline-dependent regulation of formamidopyrimidine DNA glycosylase in transgenic mice conditionally reduces oxidative DNA damage in vivo Rebecca R. Laposa,* Jeffrey T. Henderson,* and Peter G. Wells*,† *Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada; †Department of Pharmacology, University of Toronto, Toronto, Ontario, Canada Corresponding author: Peter G. Wells, Faculty of Pharmacy, University of Toronto, 19 Russell St., Toronto, Ontario, Canada M5S 2S2. E-mail:
[email protected] ABSTRACT 8-Oxo-deoxyguanosine (8-oxo-dG) is a pervasive oxidative DNA lesion formed by endogenous oxidative stress and enhanced by drugs and environmental chemicals. This lesion results in transcriptional errors and mutations and is linked to neurodegeneration, teratogenesis, cancer, and other pathologies. We demonstrate that the neonatal central nervous system of transgenic mice carrying the tetracycline-regulable DNA repair gene formamidopyrimidine DNA glycosylase (fpg) has a 50% reduction in 8-oxo-dG levels. This enhanced DNA repair is suppressed by treatment with doxycycline. For the first time, this murine model permits the level of a specific DNA oxidation product to be regulated in a temporally and spatially specific manner, allowing its role as a primary or secondary factor in neurodegenerative disease to be determined in vivo. Keywords: DNA repair • oxidative stress • brain
R
eactive oxygen species (ROS) are produced endogenously as a byproduct of normal mitochondrial respiration, and via exposure to ionizing radiation, drugs, and environmental chemicals (1, 2). Perhaps the most abundant and hazardous of these oxidative DNA lesions is 8-oxo-7,8-dihydro-deoxyguanosine (8-oxo-dG). Classically, this lesion results in G to T transversion mutations as a result of the mispairing between 8-oxodG and dA. In addition, this lesion has several important effects at the level of transcription. 8Oxo-dG causes defective transcripts as a result of base mispairing, and the 8-oxo-dG lesion reduces the rate of transcription by stalling RNA polymerase (3).
Estimates suggest that between 2 × 102 to 4 × 105 new 8-oxo-dG lesions are formed in each cell daily (4). 8-Oxo-dG is repaired with a lower efficiency than other endogenous DNA lesions (5) and is one of the most persistent oxidative DNA lesions (6), the level of which accumulates with age (7). Oxidative DNA damage has been implicated in a variety of pathophysiological processes, including neurodegenerative disease, cancer, birth defects, and the adverse effects of the normal process of aging (1, 2, 8–10). Specifically with respect to neurodegenerative disease, increased
levels of the 8-oxo-dG lesion have been observed in brains of patients with Alzheimer’s disease (AD) and spontaneous amyotrophic lateral sclerosis (ALS) (11–14). Increased levels of the 8oxo-dG lesion, but not other forms of DNA damage, have been detected in the substantia nigra of patients with Parkinson’s disease but not in the adjacent normal tissue (15). Moreover, a decrease in the activity of the enzyme 8-oxoguanine glycosylase (Ogg1), which repairs 8-oxodG, was observed in a variety of brain regions of AD patients (13). The increased levels of 8oxo-dG in these various neurodegenerative conditions may be either a primary factor in the mechanism of the disease or a secondary factor subsequent to pathologic initiation. With respect to birth defects, the potential prenatal sensitivity to 8-oxo-dG is suggested by the fact that, during development, an increase of as little as twofold in the endogenous level of 8oxo-dG is associated with increased embryonic death and teratogenesis in mice (9, 10, 16). In the case of thalidomide teratogenicity, to which rabbits are susceptible and mice are resistant, DNA oxidation is enhanced in rabbit but not mouse embryos (9). Consistent with the hypothesis that the effects of several teratogens are mediated by ROS, both the embryopathic effects and the level of DNA oxidation can be blocked by antioxidants and antioxidative enzymes and exacerbated by the inhibition of or a deficiency in antioxidative enzymes (1, 10). In particular, a specific role for DNA as a teratologically important macromolecular target is implied by the observation that p53 null and Atm null mice, which have a general deficiency in DNA repair, are more susceptible to ROS-initiating teratogens (17, 18, Laposa et al., unpublished observations). 8-Oxo-dG is repaired primarily by the base excision repair (BER) system (reviewed in ref 19). In bacterial BER, the fpg enzyme cleaves 8-oxo-dG from its sugar backbone and also excises the remaining abasic site. Subsequent enzymes in the BER pathway trim the termini and fill in the single-base gap generated by fpg. A eukaryotic functional homologue of fpg, termed oxoguanine glycosylase (ogg1), has been cloned (20–24), and the two proteins share little homology at the amino acid level and differ in their biochemical mechanisms. However, Ogg1 can functionally correct the spontaneous mutator phenotype of fpg-deficient cells (21), and vice versa (24), and fpg can protect mammalian cells from mutations initiated by ionizing radiation and chemotherapeutic agents (26–28). As the first step to determining the in vivo role that 8-oxo-dG lesions play in promoting ROSrelated diseases, we engineered binary lines of transgenic mice that overexpressed fpg, allowing for temporally and spatially regulated levels of 8-oxo-dG in various murine tissues. fpg rather than Ogg1 was chosen as the DNA repair enzyme due to the 80-fold higher activity and broader substrate range of fpg. To permit exogenous regulation of bacterial fpg expressed in mice in vivo, binary lines were generated using the “tet-off” tetracycline-dependent transactivator (tTA)/tetracycline-operator (tetO) system as described previously (29). For these studies, tTA was placed under the control of a human β-actin promoter, and fpg or β-galactosidase was placed under the control of a minimal promoter containing heptad repeats of the tet operator sequence (tetO). The binary actin-tTA:tetO-fpg murine model allows the level of a specific DNA oxidation product to be regulated in a temporally and spatially specific manner, allowing its role as a primary or secondary factor in neurodegenerative disease and birth defects to be examined in vivo.
MATERIALS AND METHODS Constructs End modifications of the fpg gene were achieved by polymerase chain reaction (PCR) of the psV2-fpg plasmid (F. Laval, INSERM) using mismatched primers containing EcoR1 sites (5′CTG CAT CTG TGA ATT CCT GGA G-3′ and 5′-CCG GAG AAT TCC CAT CAG G-3′). This was subsequently subcloned into the EcoR1 site of either the tetO plasmid pUHD10.3 (30) to generate tetO-fpg or into the EcoR1 site of an expression vector containing the CMV promoter (pcDNA3.1; Invitrogen, Carlsbad, CA) to generate the control vector CMV-fpg. For tetO-fpg, the tetO operator sequence and fpg gene were sequenced in their entirety to confirm that no errors had been introduced. CMV/tetO vectors containing Geo (β-galactosidase fused in-frame to a version of the neomycin gene) (P. Soriano, Fred Hutchinson Cancer Research Center, Seattle) were constructed in parallel to act as controls. For the expression of tTA, the CMV-tTA vector pUHD15.1 (29) was used. Cells For transient transfections, HEK 293 cells were transfected using standard calcium phosphate techniques. Transfection efficiency was assessed by β-galactosidase staining in cells transfected with Geo or cotransfected with pUHD15.1 and tetO-Geo. For stable transfections, CHO-AA8 cells (M. Rauth, Ontario Cancer Institute) were transfected by a similar technique. Cells were cotransfected with Xho1-linearized pUHD 15.1, the Xho1/HindIII fragment of tetO-fpg and EcoR1-linearized Pgk-Puro (M. McBurney, University of Ottawa), and stable clones were selected with puromycin (100 µg/ml). After the appearance of puromycin-resistant clones, individual clones were isolated and maintained as individual stable cell lines. Transgenic mice To generate tetO-fpg transgenic mice, a 1.6-kb PvuI/HindIII fragment of the tetO-fpg construct was injected into FVB/N pronuclei according to standard techniques. Three transgenic founders were identified. Founders were cross-bred with CD-1 mice to generate an outbred strain, and the transgenic offspring were identified by Southern blotting, using a 1-kB fpg probe from tetO-fpg, or by PCR, using the primers (5′-CGC CGC GGC ATA GAA CC-3′ and 5′-ATG GAT CCC CGC CGC AAA ACA-3′) specific for the fpg gene. The actin-tTA construct was produced by placing tTA downstream of the human β-actin promoter. This construct also contained 1 kB of the uncharacterized genomic region immediately 5′ to the human β-actin promoter. The 3′ untranslated region of the transgene is a modified version of the 3′ untranslated region of the human β-actin gene and was included to drive perinuclear localization of the transcript (30). Transgenic mice were engineered by standard techniques, and founders were identified by Southern blot analysis using a 1-kB tTA probe from the CMV-tTA plasmid pUHD15.1 (28). Subsequent transgenic offspring were identified by PCR using primers specific for the tTA gene (5′-CTC ACT TTT GCC CTT TAG AA-3′ and 5′-GCT GTA CGC GGA CCC ACT TT-3′), as described previously (31). The Lac1 strain of tetO-LacZ mice has been described previously (32). Transgenic tetO-LacZ mice were identified by Southern blot analysis, using a 1-kB probe
derived from the tetO-Geo plasmid, or by PCR, using primers specific for the LacZ gene (5′ACC AGC GAA ATG GAT TTT TG-3′ and 5′-AGT AAG GCG GTC GGG ATA GT-3′). β-Galactosidase staining Transiently transfected cells, 54 h posttransfection, as well as embryos at embryonic day (E) 10.5 or E 14.5, and brains from postnatal day (P) 2 or P50 mice were fixed in 0.2% glutaraldehyde for 15 min, processed according to standard procedures, and incubated with 5bromo-4-chloro-3-indoyl-β-galactoside (X-gal) for 3–24 h (31). DNA repair assay 8-Oxo-dG repair was assayed by measuring the incision of a double-stranded oligonucleotide containing a single 8-oxo-G residue, using a method modified from that of Yamamoto et al. (33). A 21-mer oligonucleotide containing an internal 8-oxoG residue (5′-CAG CCA ATC AGT [8oxoG]CA CCA TCC-3′) and its complementary sequence with C opposite 8-oxo-G (5′-GGA TGG TGC ACT GAT TGG CTG-3′) were synthesized using an 8-oxoG monomer (Glen Research, Sterling, VA). The oligonucleotide containing 8-oxoG was then 5′-end-labeled with 32 P-dATP, using T4 polynucleotide kinase (New England Biolabs, Beverly, MA), purified over a Micro Bio-Spin P-6 chromatography column (Bio-Rad, Hercules, CA), and annealed to an eightfold molar excess of its complementary strand. Cellular lysates were prepared from cells harvested 60 h after transfection (transient transfections) or 1 wk after culture in media supplemented with 1 µg/ml tetracycline or its solvent, DMSO (0.1%, v/v) (stable lines). Cells were lysed by sonication into lysis buffer (50 mM Tris-HCl [pH 7.4], 50 mM KCl, 3 mM EDTA, 5 mM magnesium acetate) and supplemented with β-mercaptoethanol (3 mM) and protease inhibitors (5 µg/ml of chymostatin, leupeptin, antipain, and pepstatin A), and the lysate was cleared of cellular debris by centrifugation, as described previously (34). A 15-µg sample of protein was incubated for 3 h at 37°C with 0.05 pmol of oligonucleotide substrate in a total reaction volume of 100 µl containing 50 mM Tris-HCl (pH 7.4), 50 mM KCl, and 500 nM EDTA. A 5-µl aliquot of the reaction solution was separated by 20% denaturing polyacrylamide gel electrophoresis (PAGE). Bands were quantified radiometrically using an Ambis 4000 scanner and One-D-Scan software (Scanalytics, Billerica, MA). The data are expressed as a percentage of the two low molecular weight bands (representing oligonucleotides from which 8oxo-dG has been excised) relative to the sum of these bands and the high molecular weight band (representing the uncut substrate). Analysis of 8-oxo-dG by high-performance liquid chromatography with electrochemical detection (HPLC-EC) The cerebral cortex and hippocampi of P2 mice derived from crosses of actin-tTA and tetO-fpg mice were snap-frozen in liquid nitrogen and stored at –80°C until use. 8-Oxo-dG content was measured by HPLC-EC according to published methods (10). For each batch, results were expressed as the ratio of fmol 8-oxo-dG/µg DNA in each sample. Batches were pooled, and due to the commonly observed differences in the absolute values of fmol 8-oxo-dG/µg DNA ratio, the relative values were expressed as a percentage of the mean value for the actin-tTA alone group.
Regulation by doxycycline Doxycycline (2 mg/ml) was dissolved in 5% sucrose in tap water, filter-sterilized, and protected from light. Animals were allowed access to water ad libitum and were administered doxycycline in the drinking water from the time of mating onward. The drinking water was changed every 3 days. For P0 and P1, newborns were injected s.c. each day with 25 mg/kg doxycycline hydrochloride in phosphate-buffered saline (PBS) (pH 7.2). RESULTS Fpg-mediated 8-oxo-dG repair activity in transiently transfected cells HEK293 cells were first transiently transfected with fpg under the control of the CMV promoter and assayed for DNA repair activity, using a double-stranded oligonucleotide containing a single internal 8-oxo-dG residue as a substrate. A representative gel is shown in Figure 1A, and the average of three experiments is shown in Figure 1B. A 24-fold increase in repair activity was observed relative to controls in the reverse orientation (Fig. 1B), indicating that fpg can be expressed and is functional in mammalian cells. Cells were subsequently transfected with CMVtTA and tetO-fpg. DNA repair activity was found to be 30-fold higher in cells cotransfected with CMV-tTA and tetO-fpg (forward) than in cells cotransfected with CMV-tTA and tetO-fpg (reverse). Cells cotransfected with CMV-tTA and tetO-fpg showed a similar DNA repair activity compared with cells transfected with CMV-fpg alone. These results indicate that the binary tetracycline system expressed fpg at a level comparable to that observed with the CMV promoter alone, at least in culture, as shown in Figure 1B. 8-Oxo-dG repair can be regulated by tetracycline in stable cell lines CHO-AA8 cells were stably cotransfected with CMV-tTA, tetO-fpg, and the puromycin resistance plasmid Pgk-Puro. After selection in puromycin, the presence of fpg in resultant clones was confirmed by Southern blotting. Individual cell lines were then cultured in the presence of tetracycline (1 µg/ml) or its solvent control (0.1% v/v DMSO) in order to examine their tTA-mediated expression of fpg. When DMSO-treated cell lysates were assayed for DNA repair activity, 8-oxo-dG repair activity was up to ninefold higher in stable cell lines relative to the parental line (Fig. 2A, 2B). When these cell lines were cultured in the presence of tetracycline at a concentration previously demonstrated to maximally inhibit tTA-mediated transcriptional activation in cell culture (34), repair activity was reduced by 80% (Fig. 2B). The bacterial fpg enzyme possesses an apyrimidinic/apurinic (AP) lyase activity that catalyzes successive β and δ elimination reactions (in our experiment, visualized on the membrane as the two lower bands) that converts an AP site to a single-strand break. (35). In contrast, the mammalian Ogg1 enzyme carries out only the β elimination reaction (36). In our experiment, this biochemical activity is visualized as a single band, the higher of the two produced by fpg. Thus, the disappearance of the lower band of the pair by the addition of tetracycline to stable clones is an indication of the disappearance of fpg-mediated, rather than Ogg1-mediated oligonucleotide excision. Some leakiness of the tetO-mediated transgene expression was observed, as described previously (37). This study demonstrated that bacterial fpg-mediated DNA repair could be effectively regulated with tetracycline in a mammalian system.
Generation of tetO-fpg transgenic mice When preparing the tetO-fpg plasmid for microinjection, the plasmid was cut at the PvuI and HindIII sites (rather than at the XhoI site, where it was cut for the stable cell lines) to allow the inclusion of an extra 600 bp of vector DNA sequence 5′ to the tetO region to protect the tetO area from exonucleases. Three transgenic founders were identified as carrying the fpg gene, as determined by Southern blotting of founders and their offspring. TetO-fpg transgenic mice were mated to lines of actin-tTA transgenic mice, and the proportion of the binary actin-tTA:tetO-fpg mice did not differ significantly from the expected Mendelian ratio, indicating no lethality in binary transgenic animals. tTA expression in actin-tTA transgenic mice To test the temporal and spatial localization of tTA, transgenic actin-tTA mice were crossed to tetO-LacZ reporter mice (Lac1 line, [34]), and resulting embryos/offspring were stained for βgalactosidase activity as shown in Figure 3. Prenatal expression was minimal (Fig. 3B, 3D), as was staining in adult (P50) animals (Fig. 3O). In contrast, tTA expression was much more widespread in neonatal brain (P2) of binary transgenic actin-tTA:tetO-LacZ mice and was high throughout the neocortex (Fig. 3G, 3J). Low levels of cortical staining were also observed in tetO-LacZ animals (Fig. 3F, 3I). However, this was much lower than that observed in binary transgenic actin-tTA:tetO-LacZ animals (Fig. 3G, 3J). As shown in Figure 3, the staining observed within the cerebellum of these animals was largely due to leakiness of the tetO-LacZ transgene. No tTA expression (as monitored by β-galactosidase activity) was observed outside the central nervous system of the neonate or adult (data not shown). To determine whether expression of the LacZ reporter gene could be regulated in this system in vivo through the application of doxycycline, we gave pregnant females doxycycline in drinking water throughout the period of gestation. In addition, newborns were injected s.c. with doxycycline (25 mg/kg) on P0 and P1, and pups were analyzed for their pattern of βgalactosidase activity on P2. Binary actin-tTA:tetO-LacZ pups treated with doxycycline (Fig. 3M) exhibited a reduction in cortical β-galactosidase activity, which was similar to that observed in doxycycline-treated littermates containing tetO-LacZ alone (Fig. 3L), implying that the tTAmediated expression of β-galactosidase is effectively inhibited by this regimen of doxycycline. Because the highest level of β-galactosidase expression was observed within the neonatal cortex, this tissue was chosen for the subsequent analysis of DNA repair activity in binary transgenic actin-tTA:tetO-fpg mice. Doxycycline regulates the endogenous level of 8-oxo-dG in vivo in binary transgenic actintTA:tetO-fpg mice To determine the efficacy of transgenic fpg to enhance DNA repair in vivo, we determined the endogenous level of 8-oxo-dG in binary transgenic actin-tTA:tetO-fpg mice as well as in littermates containing the tetO-fpg or actin-tTA transgene alone. The level of 8-oxo-dG in P2 cortex and hippocampus was analyzed by HPLC-EC and was found to be reduced by 55% in actin-tTA:tetO-fpg mice compared with control littermates (Fig. 3P) (P
